yade-dev team mailing list archive

[Bruno Chareyre <bruno.chareyre@xxxxxxxxxxx>]

> You can ask them if the interpretation and implication is right.
Many thanks. I'll try.

> > vtk doesn't have regular delaunay (triangulation of spheres - not 
> > points). So, it can be usefull for something else perhaps but it 
> > would not replace CGAL.
> I am not familiar with vtk so much and to be honest I don't know what 
> regular delaunay is ;-), but I will check it tomorrow.
I checked already, vtk don't do that. Delaunay is very common, it is a 
triangulation of point sets. Regular Delaunay is based on weighted 
points (for us : center+radius), and the only library (that I know) 
doing that is CGAL. The dual graph of Delaunay is Voronoi tesselation, 
the dual of "regular" is Dirichlet tesselation (so, "Dirichlet" is a 
better keword for search engines than "regular delaunay").


> I think there is a reason e.g. vtk is used so much compared to cgal 
> (although their focus is quite different), and that's the license. 
> Anyone should be reluctant to use something that excludes commercial 
> use, since it is commercial funding AFAIK that drives non-negligible 
> part of applied research.
I agree. And from what I've read on CGAL website, the reason for QPL 
license is precisely to let CGAL developpers get commercial
fundings. True : they don't mention commercial use from others...

> It also implies that the library is not packaged or is in special 
> repositories, whereas more standard things can be found at any decent 
> distribution -- this maintenance burden counts, for me at least, as 
> well (recall wm3 and qglviewer).
CGAL is a debian package you know?

Possible solution if we find license conflicts : I can easily split the 
code into Yade engines using triangulations, and code doing the 
triangulation that I would remove from Yade (TesselationWrapper class). 
The Yade part would be compiled only if TesselationWrapper lib is found. 
There would be only references to TesselationWrapper in Yade, and this 
is not a CGAL class (one could even write another wrapper based on a 
different library).
What do you think of this option?

Bruno


--

_______________
Chareyre Bruno
Maître de Conférences

Grenoble INP
Laboratoire 3SR - bureau E145
BP 53 - 38041, Grenoble cedex 9 - France
Tél : 33 4 56 52 86 21
Fax : 33 4 76 82 70 43
________________